Launching today
MultiCC
Turn coding agents into your own Super Agent
0 followers
Turn coding agents into your own Super Agent
0 followers
MultiCC brings Claude Code, Codex and other agents into one self-hosted workspace: task memory, delegation, tools and reviewable results. Run background work, collect outputs, and step in from your phone when a task needs a human.
Free
Launch Team
Atlassian LoomShow, don't just tell with Loom
Promoted
Hi Product Hunt, I'm Lsjwzh, the maker of MultiCC.
I wanted the coding agents I already use to work toward larger goals with a shared way to organize tasks, find earlier decisions, use tools, and bring results back. That is what I mean by a practical Super Agent workspace.
MultiCC brings those pieces together around Claude Code, Codex and other supported engines:
Start with a goal. Work in a task directly, or ask the Commander to coordinate and delegate independent work.
Keep relevant context. Tasks retain their history; scoped memory and on-demand retrieval bring earlier decisions back with their sources.
Act and wait. Use configured browser, desktop and service tools, run background jobs, and receive results through asynchronous dispatch or registered waits.
Inspect the result. Collect reports, files, previews and code diffs in the workspace. Independent task workspaces isolate parallel changes before review.
Human assistance is a supporting capability: when an agent needs a decision, sign-in or desktop action, it can ask you in the task. On a supported host, open its screen from your phone, complete the step and reply. The agent then checks the state and continues.
MultiCC is MIT-licensed and self-hosted. Bring your CLI accounts or model access; provider fees may apply. Your host and MultiCC service need to remain running. Tasks in the same conversation shell share a worktree and run serially; use independent tasks for parallel code changes. Desktop takeover currently supports macOS and Linux X11 with the required agent and permissions.
The gallery shows the real web interface with isolated example data, not a recorded agent run. Setup and current boundaries are in the repository.
What would you want your own Super Agent to take from a goal to a finished result?